SEMANTIC WEB SUPPORT FOR BUSINESS PROCESSES

Airi Salminen, Maiju Virtanen

University of Jyväskylä, PL 35 (Agora), Jyväskylä, Finland

Keywords: Business processes, semantic web, RDF, RDF Schema

Abstract: Development of semantic web technolog

ies has been initiated to improve the utilization of web resources

particularly by software applications. Semantic web is intended to extend the current web by metadata

adding meaning to web resources. In an interorganizational business process context, semantic web could be

an extension of the current intranet, extranet, and internet resources better enabling computers and people in

business processes to work in cooperation. In the paper we will explore the possibilities of the semantic web

technologies to support business processes. Particularly we will evaluate the possibilities and problems

related to the utilization of RDF (Resource Description Framework), a method supporting the formal

presentation of metadata and metadata schemas. We will use the Finnish legislative process as a case to

demonstrate the issues discussed.

1 INTRODUCTION

The wide adoption of the information and com-

munication technology innovations and web tech-

nologies from 1990’s has lead into a situation where

a number of software applications and enormous

data resources are available in business processes.

Utilization of the resources in the processes, how-

ever, still requires vastly human work.

The World Wide Web Consortium (W3C) has

itiated the development of semantic web to im-

prove the utilization of web resources. The semantic

web is intended to be “an extension of the current

web in which information is given well-defined

meaning, better enabling computers and people to

work in cooperation” (Berners-Lee, Hendler,

Lassila, 2001). The well-defined meaning is added

to the web by means of metadata. The metadata is

information about resources either accessible or

identifiable on the web. The metadata is given in a

formal, standardized format, readable and interpret-

able by software. The Extensible Markup Language

(XML) and the Resource Description Framework

(RDF) create the basis for the standard formats

(Bray, Paoli, Sperberg-McQueen, Maler & Yergeau,

2004; Manola & Miller, 2004). Formal presentation

of the metadata can be used to facilitate automated

reasoning about the meaning and trustworthiness of

resources. Ontologies are used to express semantic

metadata. An ontology defines formally the concepts

and their relationships in an application domain

(Gruninger & Lee, 2002; Klein, 2002).

In the paper our goal is to explore the possibili-

ties o

f the semantic web technologies to support

business processes. Particularly we will evaluate the

possibilities and problems related to the utilization

of RDF. By the term business process we refer to

work processes in all kinds of organizations, in

companies, public sector organizations, as well as in

other types of organizations. In the paper we will use

the legislative process as a case to demonstrate the

issues discussed.

The Finnish legislative process is an example of

a co

mplex interorganizational process participated

by many organizations, among them the Govern-

ment, the ministry on the domain of the law, the

Ministry of Justice, the Chancellor of Justice, the

Parliament of Finland, Special Committees of the

Parliament, and the President of the Republic. The

time for developing a new law may take from a

couple of months to several years. A number of

intranet systems, extranet systems, and various soft-

ware applications are used during the process. Het-

erogeneity of systems in the participating organiza-

tions and lack of communication between software

applications cause lot of extra work in the process.

The major requirements for the adoption of new

information and communication technology solu-

tions include their capability to support

•

data integration and interoperability of systems

in the process, and

468

Salminen A. and Virtanen M. (2005).

SEMANTIC WEB SUPPORT FOR BUSINESS PROCESSES.

In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 468-473

DOI: 10.5220/0002525004680473

 SciTePress

• building of intelligent services to the heterogen-

ious group of users.

Semantic web technologies seem to provide, at least

in principle, tools for the needs.

The rest of the paper is organized as follows.

Section 2 first gives an overview of our ideas about

semantic web in the context of a business process.

Section 3 will then briefly

describe RDF and RDF

Schema.

The opportunities offered by RDF/RDFS

and problems related to the definition of RDF sche-

mas will be discussed in Sections 4 and 5, respec-

tively. Before the concluding section implications of

the study for building semantic webs for business

processes will be discussed.

2 TOWARDS SEMANTIC WEB TO

SUPPORT BUSINESS

PROCESSES

A semantic web supporting business processes

involving a group of organizations would be an

extension of their current intranet, extranet, and

internet resources. A metadata standard should be

developed to express the well-defined meaning of

information used and produced in the processes.

Adding well-defined meaning to the information in a

standardized form is a challenging task.

In a business process context, the meanings are

created and shared by the human actors in the proc-

ess. There are meanings, for example, in the words

and expressions of the texts in documents, in figures,

in the organizational culture, in the gestures of peo-

ple, in the ways the activities proceed, coded in

software, and built in hardware.

To define formally the shared concepts and the

actual metadata descriptions in a business process, a

flexible description language is needed. It should

facilitate description of the common concepts related

to the process context of resources by means of

contextual metadata, the description of the concepts

related to the meaning of the resources by means of

semantic metadata, as well as description of the

physical or logical structure of the resources by

means of structural metadata. The importance of

metadata about the process context of recorded data

is pointed out also in the ISO standard 15489-1

(2001) and in the draft ISO/PDTS 23081.

Contextual and structural metadata of different

processes can be described with same concepts. In

the legislative environment, for example, the legis-

lative process for preparing a criminal law can be

described to a great extent with the same concepts as

the process for preparing a new traffic law. Also the

document structures (structural metadata) in the

processes are for the major documents the same.

Concepts for describing semantic metadata of the

documents produced in the processes however dif-

fer.

To be able to identify appropriate concepts shared

in a community in a business process, the process

and its components have to be analyzed and

described. Methods for analyzing information man-

agement in business processes, particularly for

document-centric environments where major portion

of the information created in the process activities is

recorded in documents have been described in

Salminen (2003).

Figure 1 describes the information flow in a

business process environment and at the same time a

metamodel we have used to gather the concepts

related to business processes in an environment. The

oval represents the activities of the process, the rec-

tangles three types of information repositories: ac-

tors, content items, and systems. An activity is a set

of actions performed by one or more actors in a

process. Actors are the performers of activities. An

actor is an organization, a person, or a software

agent representing a person or an organization in the

activities. Systems consist of the hardware and soft-

ware applications used to support the performance of

activities.

Figure 1: Information flows in the activities of a business process.

SEMANTIC WEB SUPPORT FOR BUSINESS PROCESSES

469

In Figure 1 content items consist of stored data

produced and used in the activities by actors. To be

able to handle metadata as resources in the same

way as other recorded content items, the metadata

should also be recorded as content items. Therefore

the content items are divided in the figure into two

types, primary content items and metadata content

items. Information is depicted by the dashed arrows.

Information needed and produced during activities is

stored in documents and other content items, in the

heads and experience of people, in the organiza-

tional culture, and in systems.

Currently we are investigating the possibilities of

RDF to be used as the description language for the

concepts of a process environment. In the following

section we will briefly describe RDF and RDF

Schema. Then we will evaluate the opportunities

offered by RDF/RDFS and problems related to the

definition of RDF schemas and schema concepts.

3 RDF AND RDF SCHEMA

RDF is a model for describing metadata about

resources. According to the specification (Manola &

Miller, 2004), a resource is anything that can be

identified on the web. As the mechanism to identify

resources RDF uses URI

references.

RDF describes metadata as simple (resource,

property, value) triples. The triples are called state-

ments. The RDF specification defines a graphical

representation for RDF descriptions as graphs, and a

textual representation using XML syntax.

RDF Schema is a language for defining vocabu-

laries intended for use in RDF statements. RDF

Schema vocabulary consists of classes, subclasses

and properties, which can be used to define compli-

cated term hierarchies. Properties are defined with

range and domain qualifiers. For example, a prop-

erty creator can be defined with a class Document as

a domain value and a class Person as a range value.

Below an example of an RDF process schema is

given. The example is a part of a schema describing

the Finnish legislative process. The schema includes

classes LegislationProject, LegislativeDocument and

ProcessPhase. Properties name, identifier, date,

createdInPhase and createdDuringLegislationProject

apply to the class LegislativeDocument, as can be

seen in the domain definition. Two last properties

have range definitions, which indicate to which class

the value of the property should belong.

<?xml version='1.0' encoding='UTF-8'?><!DOCTYPE

rdf:RDF [<!ENTITY example

"http://www.legislationexample/rdf#">]>

<rdf:RDF

xmlns:rdf=”http://www.w3c.org/1999/02/22-rdf-

syntax-ns#”

xmlns:rdfs="http://www.w3.org/2000/01/rdf-

schema#">

<rdfs:Class

rdf:about="&example;LegislationProject"/>

<rdfs:Class

rdf:about="&example;LegislativeDocument"/>

<rdfs:Class rdf:about="&example;ProcessPhase"/>

<rdf:Property rdf:about="&example;name">

<rdfs:domain

rdf:resource="&example;LegislativeDocument"/>

</rdf:Property>

<rdf:Property rdf:about="&example;identifier">

<rdfs:domain rdf:resource="&example;

LegislativeDocument"/>

</rdf:Property>

<rdf:Property rdf:about="&example;date">

<rdfs:domain rdf:resource="&example;

LegislativeDocument"/>

</rdf:Property>

<rdf:Property

rdf:about="&example;createdInPhase">

<rdfs:domain rdf:resource="&example;

LegislativeDocument"/>

<rdfs:range

rdf:resource="&example;ProcessPhase"/>

</rdf:Property>

<rdf:Property

rdf:about="&example;createdDuringLegislationPro

ject">

<rdfs:domain

rdf:resource="&example;LegislativeDocument"/>

<rdfs:range

rdf:resource="&example;LegislationProject"/>

</rdf:Property>

</rdf:RDF>

Figure 2 demonstrates the design of the process

schema using Protégé ontology editor (available at

http://protege.stanford.edu/). The left pane shows the

RDF classes of the process schema. The process

phases have been described by the class

ProcessPhase and its four subclasses. In addition to

the classes in the textual schema, there are classes

for a person and for two organizational actors min-

istry and committee. In the middle pane there are the

instances of the schema (the actual RDF descrip-

tions). The pane on the right shows the properties

used to describe a legislative document instance of

the class LegislativeDocument.

ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

470

Figure 2: Designing an RDF process schema with Protégé ontology editor.

RDF Schema offers a simple mechanism to de-

fine ontologies (Klein, 2002; Volz, Oberle & Studer,

4 OPPORTUNITIES IN RDF

2003). Compared to some other ontology languages,

RDF is however limited. The major limitation is the

lack of reasoning support, which RDF does not in-

clude. To get more capabilities for describing re-

sources, RDF Schema should be developed further

The XML syntax of RDF makes RDF descrip-

tions machine-readable. The XML syntax brings

several well-known advantages, such as system

independence and possibility to be used as a format

in the data interchange of various systems.

or its use should be extended with richer schema

languages. According to Manola and Miller (2004)

useful additional capabilities would be:

• cardinality constraints,

• adding transitivity definition to a property,

• adding a definition that a property is a unique

identifier,

• definition of two classes or instances represent-

ing the same class or individuals, respectively,

• definition of constraints on the range or cardinal-

ity of a property that depend on the class (e.g. the

range of the property identifier differs if the de-

scribed class is LegislativeDocument or Legisla-

tionProject), and

• ability to describe new classes in terms of combi-

nations (e.g. unions and intersections) of other

classes, or to define that two classes are disjoint

(e.g. no resource is an instance of both Legisla-

tiveDocument and LegislationProject classes).

RDF offers many kinds of opportunities for de-

scribing metadata for a business process. In this

section we evaluate these opportunities.

RDF together with RDF Schema supports the

description of various types of metadata including

semantic, contextual and structural metadata. Differ-

ent RDF descriptions can be attached to a single

resource. Thus a collection of resources can be

described for different user groups by different

terms. In an inter-organizational environment there

can be many groups of actors with varying terms for

the same concepts.

For example, in a legislative

environment

citizens and lawyers could have their

own schemas

In complex interorganizational processes the

metadata schemas have to be created gradually. RDF

supports that kind of proceeding, because RDF

schemas

can be flexibly extended and also merged

with other vocabularies from various sources with

SEMANTIC WEB SUPPORT FOR BUSINESS PROCESSES

471

the help of XML namespaces (Hunter & Lagoze,

2001; Candan, Liu & Suvarna 2001).

The RDF/RDFS has remarkable flexibility com-

pared to other modeling approaches. Instead of de-

scribing classes with certain properties, RDF

Schema

introduces properties as first-class entities

like classes

. Properties apply to specific classes

according to domain and range specifications.

As an ontology language RDFS facilitates buid-

ing

intelligent semantic web services, such as se-

mantic browsing service and semantic recommenda-

tion (Hyvönen & al., 2004; Quan and Karger, 2004).

Semantic recommendation means that additional

related links are shown to the user parallel to the

original search result. According to Middleton,

Alani, Shadbolt and De Roure (2002), people often

feel it difficult to express what they want or what

they are looking for. Semantic recommendation

might be a useful tool for information retrieval in

business process environments.

5 CHALLENGES IN DESIGNING

RDF SCHEMAS

Noy and McGuinness (2001) state that probably the

most difficult task in the schema or ontology

design is to find

consensus between people in the

community. People may have completely different

views about important and central terms in the or-

ganization and on the domain. In a community of

several organizations achieving consensus may be

impossible. There are cases where someone should

be in a position to make decisions on the schema

concepts when representatives of the community do

not reach agreement. Identifying the owner of the

process environment might solve the problem, but

getting agreement of the owner is sometimes ex-

tremely difficult in an interorganizational process. In

the case of the Finnish legislative process no process

owner has been identified. The needs for standardi-

zation and integration throughout the process have

however initiated discussions about the possible

owner.

Heterogeneity of users in an organization is

also a challenge in schema design. Heterogeneity

concerns, for example, the roles of people in work

processes and the language they are accustomed to

use in communication.

Schema design experts hardly are experts of the

target business process. The schema designers have

to do very close cooperation with users to learn the

concepts and needs of the environment. It may how-

ever be hard to find from a large set of actors the

users having time to collaborate and having enough

expertise on the area. If the users have no knowledge

about metadata or ontologies, it is sometimes diffi-

cult

to motivate them to spend their time in the de-

sign work.

Ontologies introduced for semantic web usually

describe rather static resources like, for example,

museum items in the MuseumFinland service (Hy-

vönen & al., 2004). The items themselves do not

change, only the meanings given by people may

change. The nature of processes and also the nature

of documents created in the processes is dynamic

and causes thus more frequent update needs for the

schema. Also according to the draft ISO/PDTS

23081 metadata schemas

need to be continuously

updated

to reflect changes in an organization and in

the business.

In many business processes the content of data

resources varies so much that the development of

content ontologies with detailed semantic metadata

covering those resources is extremely difficult. The

number of concepts easily increases over the limit to

be systematically updated. In the legislative envi-

ronment, for example, the legal terminology is huge

and the current legal terminology in Finland is

closely related to the EU legal terminology. Rela-

tions between various laws

are continuously chang-

ing

and totally new areas evolve to be regulated by

laws (such as gene technology).

6 IMPLICATIONS

Our analysis has clearly shown that there are in-

teresting opportunities in using RDF to support in-

formation management in business processes but the

problems for creating the RDF schemas and de-

scriptions are extremely difficult, especially in the

cases of complex interorganizational processes.

These are however also the cases where the needs

for new solutions are most urgent and even minor

enrichment of the process environment by well-

designed metadata might serve the work in the proc-

ess. Thus we argue that careful analysis to identify

the most essential core metadata would be important

in business process environments.

Our experience and analysis has shown the diffi-

culty of developing ontologies to describe the

meaning of content items when the content concerns

various domains. The number of concepts poten-

tially needed in the ontology may be huge and find-

ing agreement about the standard ontology may be

extremely difficult. In ontologies describing the

process context of the content items, the number of

concepts needed is evidently smaller.

In developing process schemas it is important to

look for possible existing standard terminologies.

For example, in spite of the many changes taken

ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

472

place during the last decades in the Finnish

legislative work, there still are stable terms for a

number of central concepts. The terms are described

in a published dictionary (Torniainen, 1999).

Communication between schema designers and

people representing the organizations involved is

important during the schema design. Views of

various organizations should be taken into account.

Coordination of the collaboration is essential to

facilitate participation of all interested parties.

7 CONCLUSION

In interorganizational business processes a major

problem is the heterogeneity of software

applications and lack of interoperability. Semantic

web technologies offer, at least in principle, a

solution to the problem. Building semantic web

solutions essentially means developing standardized

metadata solutions. In the paper we discussed the

metadata types needed in business processes. RDF

is the model intended to be used for semantic web

metadata. In the paper we analyzed the possibilities

to use RDF to support information management in

business processes and the challenges in designing

schemas for the purpose. In spite of the major

problems in the schema design and maintenace of

ontologies, the use of RDF seems to offer interesting

possiblities. Therefore building and evaluating

experimental solutions in real business process

environments will be important. Our future work

will include designing and evaluating experimental

RDF schemas and descriptions for the legislative

process.

REFERENCES

Berners-Lee T., Hendler J. & Lassila O. 2001. The

Semantic Web. Scientific American 284(5), 34-43.

Retrieved July 15, 2004, from

http://www.sciam.com/article.cfm?articleID=0004814

4-10D2-1C70-84A9809EC588EF21

Bray T., Paoli J., Sperberg-McQueen C., Maler E. &

Yergeau F. 2004. Extensible Markup Language

(XML) 1.0 (Third Edition). W3C Recommendation.

Retrieved from

http://www.w3.org/TR/2004/REC-xml-20040204/

Brickley D. & Guha R.V. (Eds.) 2004. RDF Vocabulary

Description Language 1.0: RDF Schema. W3C

Recommendation Retrieved June 17,2004, from

http://www.w3.org/TR/rdf-schema/.

Candan K., Liu H. & Suvarna R. 2001. Resource

Description Framework: Metadata and its

applications [Electronic version]. ACM SIGKDD

Explorations Newsletter 3(1), 6-19.

Gruninger M. & Lee J. 2002. Ontology applications and

design [Electronic version]. Communications of the

ACM 45(2), 39-41.

Hunter J. & Lagoze C. 2001. Combining RDF and XML

schemas to enhance interoperability between metadata

application profiles. In Proceedings of Tenth

International Conference on World Wide Web, Hong

Kong, 457-466.

Hyvönen E., Saarela S., Viljanen K., Mäkelä E., Valo A.,

Salminen M., Kettula S. & Junnila M. 2004. A portal

for publishing museum collections on the Web. In

Proceedings of ECAI/PAIS 2004, Valencia, Spain

(forthcoming).

ISO 15489-1. 2001. International Standard. Information

and documentation – Records management. Part 1:

General.

ISO/PDTS 23081. 2003. International Standard / Proposed

Draft Technical Specification. Information and

documentation – Records Management Processes –

Metadata Records. Part1: Principles.

Klein M. 2002. Interpreting documents via an RDF

Schema ontology. In Proceedings of the

13th International Workshop on Database and Expert

Systems Applications, Aix-en-Provece, France, 1-5.

Manola F. & Miller E. (Eds.) 2004. RDF Primer. W3C

Recommendation. Retrieved from

http://www.w3.org/TR/2004/REC-rdf-primer-

20040210/

Middleton S., Alani H., Shadbolt N. & De Roure D. 2002.

Exploiting synergy between ontologies

and recommender systems. In Proceedings of Semantic

Web Workshop 2002 at the eleventh International

World Wide Web Conference, Hawaii, USA.

Noy N. & McGuinness D. 2001. Ontology development

101: A guide to creating your first

ontology. Retrieved June 14,2004, from KSL (Knowledge

Systems Laboratory) of Stanford University Web site:

http://protege.stanford.edu/publications/ontology_deve

lopment/ontology101.pdf

Quan D. & Karger R. 2004. How to make a Semantic Web

browser. In Proceedings of the 13th Conference on

World Wide Web, New York, USA, 255-265.

Salminen A. 2003. Towards digital government by XML

standardization: Methods and Experiences. In

Proceedings of the XML Finland 2003, Kuopio,

Finland, 5-15.

Torniainen P. (Ed.) 1999. Valtioneuvoston sanasto. Edita.

Helsinki.

Volz R., Oberle D. & Studer R. 2003. Views for light-

weight Web ontologies. In Proceedings

of the 2003 ACM Symposium on Applied Computing

(SAC), Melbourne, FL, USA, 1168-1173.

SEMANTIC WEB SUPPORT FOR BUSINESS PROCESSES

473